consider adding a `nid()` function to handle numeric ids

[jefferis]

ids are identifiers not indices. Something like this might be possible:

nid <- function(...) {
  sc <- sys.call()
  scc=as.character(sc)
  stopifnot(scc[1]=='nid')
  i64=bit64::as.integer64(scc[-1])
  max64=bit64::as.integer64("9223372036854775807")
  if(isTRUE(any(i64<0, na.rm = T)) || 
     isTRUE(any(i64==max64, na.rm = T))) {
    badids=!fafbseg::valid_id(i64)
    if(any(badids)) {
      warning("there are bad ids:", paste(scc[-1][badids], collapse = ','))
      i64[badids]=NA
    }
  }
  class(i64)=union("nid", class(i64))
  i64
}

It doesn't look like there is a conflicting nid definition in packages that natverse users typically have. This might even belong in nat.utils. Would also need to figure out how to handle arguments that contain numeric ids rather than actually being one. This could be part of the solution to the common problem of indexing a neuronlist with numeric ids.

[jefferis]

A note that while it is possible to intercept numbers entered into source code files before they are converted to numeric doubles, the same is not true for the console. Therefore I don't think it will ever be possible to have an nid function that works with 64 bit integers in the console – until such a time as R has native support for 64 bit ints ...